基于GM(1,1)模型的五指山市林地景观动态模拟

Dynamic simulation of woodland scenes in Wuzhishan city based on GM(1.1) model

对海南省五指山市1998年、2003年、2007年3期遥感影像数据进行解译,统计针叶类、硬阔类、软阔类、阔叶混交林、竹林类、(乔)经济林、(灌)经济树种、其他灌木类、非林地等景观类型的面积,运用MATLAB建立各类型的预测模型,并分别对2012、2016、2021和2025年进行预测。结果表明:(1)各景观类型预测模型平均相对误差均小于0.01,精度检验处于Ⅰ级(优),各模型的相关系数均小于0.3,可用于中长期预测;(2)到2025年,林地面积为102 693.86 hm2,非林地面积为12 800.74 hm2;(3)林地中各景观类型按面积大小排序为:阔叶混交类>乔木经济林类>硬阔类>针叶类>软阔类>竹林类>灌木经济林类>其他灌木类。

The landscape areas of forest landscapes in Wuzhishan City of Fujian province in 2012, 2016, 2021 and 2025 were predicted by interpreting the remote sensing images in 1998, 2003, 2007; the and forest areas of conifers, hardwood class, soft broad categories, broadleaf forest, bamboo forest type, （Joe） forest, （irrigation） economic trees, other shrubs, and other landscape types of non-forest area were counted respectively; and then the forecasting models for various type forests in Wuzhishan city were established by using MATLAB. The results show that （1） Each landscape type prediction model mean relative error was less than 0.01, the testing accuracy were all in grade I （excellent）, the correlation coefficients of each model were less than 0.3, which can be used to medium- and long- term forecasting;（2） there will be 102,693.86 hectares of forest and non-forest area of 12,800.74 hectares by 2025;（3） the forest landscape types ordered according to the size from big to small： broad-leaved mixed class, the class tree forest, hardwood class, conifers, broad classes of soft, bamboo class, shrub forest type, other shrubs.